Sponsor Miner
Use this skill to find companies that already sponsor developer-facing open source READMEs and may be good sponsors for a target GitHub repository.
Prerequisites:
python33.9 or newer- GitHub CLI installed and authenticated so
gh auth tokensucceeds
Workflow
- Extract the single target repo URL from the user request.
- If no repo URL is provided, ask for one.
- If multiple repo URLs are provided, ask the user to choose one.
- Run the bundled miner from the current workspace:
Resolvepython3 <this-skill-dir>/scripts/sponsor_miner.py <target_repo_url><this-skill-dir>from the path of thisSKILL.md. - Read the session path printed by the script.
- Review:
<session>/run-summary.md<session>/shortlist.csv<session>/sponsor-candidates.csvwhen evidence needs auditing
- Verify each promising lead:
- open
evidence_urland confirm it shows real sponsorship, support, or promotion - check whether the sponsor's audience overlaps with the target repo's users
- find the best public contact, sponsorship, founder, DevRel, marketing, or company contact URL
- open
- Fill
<session>/verified-leads.csv; the script creates it with headers only.
Output
Write final verified leads to:
<session>/verified-leads.csv
Use this exact schema:
company,domain,source_repo,evidence_url,fit_score,fit_notes,contact_url,status
Use these statuses:
verified: good lead and contact URL foundcontact_missing: good lead but no usable contact URL foundneeds_review: plausible lead but evidence or fit remains uncertainrejected: not a real sponsor lead or no plausible audience overlap
Rewrite fit_notes as an actual one-sentence rationale. Do not leave the raw README snippet from shortlist.csv as the final note.
Reject leads when they are only donation links, generic badges, CDN/package registry links, conference templates, academic acknowledgements, placeholders, podcast/ad-detection examples, or companies with no plausible audience overlap with the target repo.
Do not send outreach unless the user explicitly asks.